Abstract
Logistic and probit models are the most popular regression model for binary
outcomes. A simple robust alternative is the robit model, which replaces the
underlying Normal distribution in the probit model with a Student t–distribution.
The heavier tails of the t–distribution (compared with the Normal distribution)
means that model outliers are less influential. Robit regression can be fit as a
generalized linear model with the link function defined as the inverse cumulative t–
distribution function with a specified number of degrees of freedom (df), and it has
been advocated as being particularly suitable for estimating inverse–probability
weights and propensity scoring more generally. Here we describe a new package
called robit that implements robit regression in Stata.
outcomes. A simple robust alternative is the robit model, which replaces the
underlying Normal distribution in the probit model with a Student t–distribution.
The heavier tails of the t–distribution (compared with the Normal distribution)
means that model outliers are less influential. Robit regression can be fit as a
generalized linear model with the link function defined as the inverse cumulative t–
distribution function with a specified number of degrees of freedom (df), and it has
been advocated as being particularly suitable for estimating inverse–probability
weights and propensity scoring more generally. Here we describe a new package
called robit that implements robit regression in Stata.
| Original language | English |
|---|---|
| Journal | STATA JOURNAL |
| Publication status | Accepted/In press - 1 Nov 2022 |